The post-earthquake evacuation process was divided into multiple periods and the casualties were transported through ambulances and helicopters. Considering the survival probability and psychological cost changes of casualties during different periods, a multiple-period dynamic fuzzy programming model for emergency medical facility locating and casualty evacuation was established to maximize the survival probability and minimize the psychological cost of casualties. The model used a triangular fuzzy number to describe the casualty uncertainty. The model was solved using Cplex and sensitivity analysis was performed using a real case based on the Wenchuan Dujiangyan earthquake. The following conclusions were obtained. To increase the survival rate of casualties and reduce psychological costs, severely injured persons should be prioritized for evacuation in the early stages of disasters, while those with mild injuries should be prioritized for evacuation in the later stages of disasters. Transport vehicles used for emergency treatment had a greater impact on the survival rate of injury casualties than those used for subsequent treatment. Temporary hospitals with a small capacity led to a decrease in the number of casualties with mild injuries, while rear hospitals with a small capacity led to a decrease in the number of those with severe injury. Excessive vehicles did not have an impact on the survival rate of casualties and psychological costs.